Ebrahim, M. Shahid and Girma, Sourafel and Shah, M ... · M. Shahid Ebrahim Bangor University, UK...
Transcript of Ebrahim, M. Shahid and Girma, Sourafel and Shah, M ... · M. Shahid Ebrahim Bangor University, UK...
Ebrahim, M. Shahid and Girma, Sourafel and Shah, M. Eskandar and Williams, Jonathan (2014) Rationalizing the value premium in emerging markets. Journal of International Financial Markets, Institutions and Money, 29 . pp. 51-70. ISSN 1042-4431
Access from the University of Nottingham repository: http://eprints.nottingham.ac.uk/47147/1/rationalisng.pdf
Copyright and reuse:
The Nottingham ePrints service makes this work by researchers of the University of Nottingham available open access under the following conditions.
This article is made available under the University of Nottingham End User licence and may be reused according to the conditions of the licence. For more details see: http://eprints.nottingham.ac.uk/end_user_agreement.pdf
A note on versions:
The version presented here may differ from the published version or from the version of record. If you wish to cite this item you are advised to consult the publisher’s version. Please see the repository url above for details on accessing the published version and note that access may require a subscription.
For more information, please contact [email protected]
1
Rationalizing the Value Premium in Emerging Markets
M. Shahid Ebrahim
Bangor University, UK
Sourafel Girma
University of Nottingham, UK
M. Eskandar Shah
INCEIF (International Centre for Education in Islamic Finance), Malaysia
Jonathan Williams
Bangor University, UK
Correspondence Address: M. Shahid Ebrahim
Professor of Banking and Finance
Bangor Business School
Bangor University
Bangor LL57 2DG
United Kingdom
Tel: +44 (0) 1248 38 8181
Fax: +44 (0) 1248 38 3228
E-Mail: [email protected]
1
Rationalizing the Value Premium in Emerging Markets
Abstract: We reconfirm the presence of value premium in emerging markets. Using the
Brazil-Turkey-India-China (BTIC) grouping during a period of substantial economic growth
and stock market development, we attribute the premium to the investment patterns of
glamour firms. We conjecture based on empirical evidence that glamour firms hoard cash,
which delays undertaking of growth options, especially in poor economic conditions. Whilst
this helps to mitigate business risk, it lowers market valuations and drives down expected
returns. Our evidence supports arguments that the value premium is explained by economic
fundamentals rather than a risk factor that is common to all firms.
JEL Classifications: G110 (Portfolio Choice; Investment Decisions), G120 (Asset Pricing).
Keywords: Asset Pricing, Growth (i.e., Glamour) Stocks, Multifactor Models, Real
Options, Value (i.e., Unspectacular) Stocks.
2
I. Introduction
″Growth stocks, which derive market values more from growth options, must
therefore be riskier than value stocks, which derive market values more from
assets in place. Yet, historically, growth stocks earn lower average returns than
value stocks.″ (Lu Zhang, 2005, pp 67)
Fama and French’s (1992) finding, that a single factor encapsulating risk (beta) does
not adequately explain cross-sectional differences in stock returns, has motivated an
important strand of research on asset pricing, reigniting the debate on the fundamental
relationship between risk and return, and challenging the widely-accepted Capital Asset
Pricing Model (CAPM). Subsequently, numerous theoretical and empirical studies examine
the cross-sectional variation in stock returns with many finding patterns unexplained by the
CAPM and commonly known as anomalies.
This paper examines one of the most pronounced anomalies, the value premium puzzle.
Portfolios formed on the basis of high book-to-market (BE/ME), cash flow-to-price (C/P) and
earnings-to-price (E/P) are reported to earn significantly higher risk-adjusted returns than
portfolios with contrasting characteristics. However, the previous literature fails to achieve a
consensus on the source of the value premium (Chou et al., 2011). The objectives of this
paper are to confirm the presence of value premium in a new market, to provide a new
rationalization based on economic fundamentals, and to reconcile the diverging perspectives
which are apparent in the literature. The value premium reflects a tendency for ‘glamour
firms’ to hoard cash and delay implementation of growth strategies, particularly in times of
economic uncertainty (Titman, 1985; McDonald and Siegel, 1986; Ingersoll and Ross, 1992).
Since growth (glamour) stocks derive their market value from embedded growth in the form
3
of real options (Zhang, 2005), we argue that cash hoarding limits their exposure to risk but
exerts a significant detrimental impact on their stock returns.
The theoretical basis for our analysis derives from Fama and French (1995) and Daniel
and Titman (1997). Fama and French (1995) develop a three-factor model, in which the
factor that captures distress risk, known as HML, is lower for growth (glamour) firms than
for value firms. The debate centers on whether lower distress risk accounts for the
discrepancy in average returns between value firms and growth firms (Fama and French,
1995) against claims that distress risk does not contribute to the value premium (Dichev,
1998; Griffin and Lemmon, 2002). We contend that both the cash-drag factors and firm
characteristics, as highlighted by Daniel and Titman (1997), are of relevance.
In comparison with value firms, growth firms face a wider array of strategic options,
carrying various levels of risk. These firms may limit their exposure to risk by abstaining
from investing resources in risky strategies, especially in poor economic environments.
Accordingly, growth firms hoard cash when economic conditions are tough, and realize lower
returns. By contrast, value firms are prominent in mature and/or declining markets and face a
more limited range of options. Such firms face financial risk, as well as business risk, owing
to a tendency to use existing assets as collateral in order to leverage earnings. They have less
flexibility in managing their risk, because past sunk-cost investment in assets is irreversible
(Zhang, 2005). Our approach in rationalizing the value premium is consistent with the
neoclassical framework, in which low-risk assets yield lower returns and vice versa.
Our research draws on two recent studies that contrast the approach of Fama and
French (1995) with Daniel and Titman (1997). In a similar vein to Daniel and Titman (1997),
Chen et al. (2011) propose a three-factor model incorporating factors with greater
explanatory power for cross-sectional returns than the Fama and French model. We aim to
extend these findings, by obtaining results that are not sample-specific (a limitation of Chou
4
et al., 2011), and by adopting a real options framework in cases where the Net Present Value
investment perspective (Chen et al., 2011) is inapplicable.1 This paper is among the few that
try to reconcile differences not only between the neoclassical asset-pricing literature (Fama
and French versus Daniel and Titman), but also the neoclassical and behavioral literature.
Our application is to a new grouping of four, large emerging markets: namely, Brazil,
China, India and Turkey, the BTIC group.2 Each of these economies has achieved
remarkable growth since the early 2000s, which implies there are many firms endowed with
plentiful growth options.3 Our paper addresses two research questions: (i) is a statistically
significant value premium present in the BTIC? (ii) is it possible to rationalize the value
premium and reconcile the apparently conflicting views in the literature? To investigate
questions (i) and (ii) we source relevant variables from 1999 to 2009 to allow our analysis of
value anomalies to be conducted under generally favourable economic conditions including
1 Ingersoll and Ross (1992, p. 2) explain this as follows:
“If in making the investment today we lose the opportunity to take on the same project in the future, then
the project competes with itself delayed in time. In deciding to take an investment by looking at only its
NPV, the standard textbook solution tacitly assumes that doing so will in no way affect other investment
opportunities. Since a project generally competes with itself when delayed, the textbook assumption is
generally false. Notice, too, that the usual intuition concerning the “time value of money” can be quite
misleading in such situations. While it is true NPV postponing the project delays the receipt of its positive
NPV, it is not true that we are better off taking the project now rather than delaying it since delaying
postpones the investment commitment as well.
Of course, with a flat, non-stochastic yield curve we would indeed be better off taking the project now, and
this sort of paradox could not occur. But that brings up the even more interesting phenomenon that is
central focus of this article, the effect of interest-rate uncertainty on the timing of investment”.
2 Non-availability of data for Russia precludes investigation of the BRIC quartet (Brazil, Russia, India and
China). We select Turkey as an alternative large European emerging economy, and note the creator of the
BRIC acronym, Jim O’Neill, plans to include Turkey and three other emerging economies in a new
grouping (Hughes, 2011).
3 The rapid growth of emerging economies, and the impact on the world economy, is discussed extensively
in the popular press. For instance, the August 6th
, 2011 issue of the Economist magazine highlights several
noteworthy statistics on emerging versus developed economies. In 2010 emerging economies account for
nearly 54% of world Gross Domestic Product (GDP) measured at purchasing power parity and 75% of
global real GDP growth over ten years. Exports exceed 50% of the world total, and imports account for
47%. Foreign direct investment (FDI) to emerging economies significantly increased recently as have
commodity consumption and capital spending. Stock market capitalization equals 35% of the world total,
this share having tripled since 2000.
5
increasing stock market integration following the liberalization of equity markets in the
BTIC.
By way of preview, we find a significant value premium in the BTIC which is not new
for emerging markets but re-emphasizes the value premium is not a developed country
phenomenon. A second result, which is based on the widely-used Altman Z-score model,
shows value firms are no more prone to risk than growth firms, but value firms employ more
leverage. Our evidence suggests the investment patterns of growth firms are the source of the
value premium. In an asset-pricing model, the HML coefficients of growth portfolios are
small during low-growth periods and considerably larger during high-growth periods. This
pattern is consistent with the hypothesis that growth firms delay the implementation of new
strategies in periods of economic uncertainty in order to limit their risk. The HML
coefficients of growth portfolios are sensitive to changes in size (total assets). Accordingly
growth in total assets, interpreted to proxy implementation of growth strategies, explains the
change in business risk of growth firms. This affirms our hypothesis that the investment
patterns of growth firms impact significantly on their risk and return.
Our results also resolve the various perspectives in the literature in three ways. First,
Fama and French (1995) and Daniel and Titman (1997) attribute the outperformance of value
stocks to different causes. Daniel and Titman (1997) explain performance differentials
between value and growth stocks as being due to the characteristics of firms as opposed to
covariance with risk factors. Value stocks outperform because growth firms tend to hoard
cash and delay undertaking the growth options they are endowed with and this drags down
their returns. In the Fama and French (1995) framework, this phenomenon is attributed to
distress risk. Second, growth firms’ flexibility to manage their embedded growth options to
their operational and strategic advantage yields not only profit, but also provides utility in its
own right: embedded options provide ‘glamour firms’ with an allure with which to entice
6
investors. ‘Fascination’ with growth firms (Sargent, 1987) creates a premium in price (and
hence a discount in returns), which helps reconcile the neoclassical and behavioral
perspectives. Third, the over-reaction hypothesis of DeBondt and Thaler (1985 and 1987) is
rationalized through the volatile nature of value firms’ leveraged equity, which is akin to Call
options. These options are depressed in poor economic states but rebound in prices with
improving economic climate.
This paper is organized as follows. Section II reviews the relevant literature, and
identifies the research questions and methodology. Section III offers a brief synopsis of
market developments in the BTIC. Section IV presents data and methodology. The results
and checks for robustness are in Sections V and VI. Section VII offers concluding
comments.
II. Literature Review
The Capital Asset Pricing Model (CAPM) posits that risk, measured using beta,
accounts for the cross-section of expected stock returns (Sharpe, 1964; Lintner, 1965a,b;
Mossin, 1966). Numerous empirical studies test the model, on the assumption that beta is the
sole explanatory variable with a positive and linear relation to asset return, yet results are
inconclusive. Several early empirical studies (Black et al., 1972; Blume and Friend, 1973;
Fama and McBeth, 1973) provide support for the CAPM. Later studies, however, are more
critical, citing evidence of anomalies and questioning the validity of the assumptions (Roll,
1977; Basu, 1977, 1983; Stattman, 1980; Banz, 1981; DeBondt and Thaler, 1985; Rosenberg
et al., 1985; Bhandari, 1988; Jegadeesh and Titman, 1993; Cohen et al., 2002; Titman et al.,
2004). Fama and French (1992) conclude that CAPM with a single factor does not
adequately explain cross-sectional stock returns, and propose a three-factor model to capture
7
the multidimensional aspect of risk, comprising: (i) a market factor (RM-Rf); (ii) a size
premium (SMB) (return on a portfolio of small stocks minus return on a portfolio of large
stocks); and (iii) a value premium (HML) (return on a portfolio of value stocks with high
BE/ME minus return on a portfolio of growth stocks with low BE/ME). This approach builds
on Merton’s (1973) intertemporal CAPM and Ross’s (1976) arbitrage pricing theory (APT).
Fama and French (1993, 1995) show SMB and HML are related to risk factors in stock
returns; and both factors contain explanatory power for the cross-sectional variation in stock
returns.
The three-factor model has attracted a great deal of academic interest, much of it
centered on the source of the value premium. In line with the hypothesis of rational pricing,
Fama and French (1993) and Chen and Zhang (1998) suggest value firms are riskier and
more likely to be subject to financial distress than growth firms. Fama and French (1995)
demonstrate that value [growth] stocks are generally associated with persistently low [high]
earnings, creating a positive [negative] loading on HML, which implies higher [lower] risk of
distress. On the contrary, Zhang (2005) claims value firms are riskier because their assets at
risk are larger than those of growth firms. This becomes particularly evident in poor
economic environments, where firms with fixed assets pose greater risks for investors than
those with growth options. Value firms are burdened with unproductive capital that cannot
be liquidated in order to recover the cost of the original investment.
More recent value premium studies investigate how differing states of the world affect
the strength of the premium. The empirical evidence contained in these papers demonstrates
the value premium is time varying and sensitive to changes in economic conditions. Stivers
and Sun (2010) show the value premium is countercyclical and higher during periods of weak
economic fundamentals whereas Guo et al. (2009) find value stocks are riskier than growth
stocks under weak economic conditions. Similarly, expected excess returns on value stocks
8
are more sensitive to deteriorating economic conditions during episodes of high market
volatility (Gulen et al., 2011). Finally, the size of the value premium positively relates to its
conditional volatility (Li et al., 2009).
Several alternative theories also seek to explain the value premium. Focusing on
investor sentiment and trading strategies, Lakonishok et al. (1994) and Haugen (1995)
attribute a tendency for value (‘unspectacular’) firms to produce superior returns to an
irrational tendency on the part of investors to extrapolate the past strong [weak] performance
of the growth [value] firm into the future. Investors overbuy [oversell] the growth [value]
firm’s stock. Lower [higher] than expected realized performance on the part of the growth
[value] firm generates a low [high] stock return.4 Similarly De Bondt and Thaler (1985,
1987) observe that poorly performing stocks (‘losers’) over the past three-to-five years
outperform previous ‘winners’ during the subsequent three-to-five years.
Daniel and Titman (1997) claim the explanation for the value premium lies in firm
characteristics rather than covariance risk. High covariance between the returns of value
stocks reflects common firm characteristics, such as, the line of business or industry
classification. Daniel and Titman (1997) show high covariance between stock returns bears
no significant relation with the distress factor. Evidence of high covariance precedes any
signs of financial distress on the part of value firms.5
Other possible explanations for the value premium focus on methodological issues in
empirical studies. Banz and Breen (1986) and Kothari et al. (1995) suggest sample selection
may be biased towards firms that survived a period of distress, rather than those that failed.
The notion that survivorship bias accounts for the value premium is rejected, however, by
Davis (1994), Chan et al. (1995) and Cohen and Polk (1995). Data ‘snooping’, an eventual
4 La Porta et al. (1997) find value firms enjoy a systematically positive earnings surprise while glamour
firms display the opposite.
5 Lee et al. (2007) find stock characteristics better explain UK value premiums.
9
tendency for repeated testing using the same data to reveal spurious patterns is cited as an
explanation for the value premium phenomenon (Lo and MacKinlay, 1988; Black, 1993;
MacKinlay, 1995; Conrad et al, 2003). Any such tendency might be mitigated by testing data
from different periods or countries (Barber and Lyon, 1997).
In spite of contentions that the value premium is a developed country phenomenon (see
Black, 1993; MacKinlay, 1995; Campbell, 2000) other evidence supports the proposition of a
value premium in emerging markets. In samples containing some (or all) of the BTIC group,
Rouwenhorst (1999) uses a cross section analysis of returns across twenty emerging markets
between 1982 and 1997, whereas Barry et al (2002) study 35 emerging markets from 1985 to
2000. De Groot and Verschoor (2002) confirm the value premium for a sample of south east
Asian markets between 1984 and 2000. Ding et al (2005) also examine markets in south and
east Asia. Their study recognizes cross country differences with respect to the value
premium, but unfortunately the data it uses end prior to the Asian crisis of 1997, which is a
major drawback as more recent value premium studies suggest the premium may change in
an economic downturn. Given the very limited work on explaining the value premium in the
context of emerging markets, many questions surrounding the value premium remain
unanswered. This paper will test for the value premium during the most recent period of
growth in emerging markets whilst offering a cross country perspective and it will also
analyse the source of the premium in emerging markets, which few papers undertake.6
III. Market developments in the BTIC
The BTIC countries liberalized their stock markets between the late 1980s and mid-
1990s. The benefits of liberalization include higher levels of real economic growth and real
6 One exception is a study of Singapore that attributes the value premium to a one-way overreaction of value
firms (Yen et al., 2004).
10
investment (Bekaert et al., 2003a) with studies reporting a significant relationship between
stock market liquidity and economic growth (Levine and Zervos, 1998). Liberalization is
expected to increase the level of integration between emerging stock markets and the world
market which should lower country risk premiums. In the 1990s emerging markets
represented a new asset class and stock prices in those markets were driven upwards by
investors seeking to diversify their portfolios, leading to permanently lower costs of capital in
the emerging markets (Bekaert and Harvey, 2003). Several studies propose methods to
effectively date when liberalization takes place (see, Bekaert et al, 2003b; Kim and Kenny,
2007).7 We calculate several indicators to proxy stock market development and growth in the
BTIC (Source: World Development Indicators). In 2010, the BTIC share of global stock
market capitalization equals 15.10% (US$ 8 trillion) increasing by nearly five-fold since the
2000 level of 3.18%. The ratio of listed firms’ market capitalization-to-GDP measures the
level of stock market deepening and it shows the combined BTIC stock market is nearly two
times deeper in 2010 compared to 2000 (78.47% c.f. 39.66%). Similarly, the BTIC stock
markets are more liquid as measured by the ratio of value of stocks traded-to-GDP. Taking
1992 as a year to represent pre-stock market liberalization, liquidity increases by a factor of
eight in Brazil (42.09% in 2010); over thirty four in China (135.40% in 2010); almost nine in
India (62.74% in 2010); and over eleven in Turkey (57.66% in 2010).
IV. Data and Methodology
We source data on listed firms in Brazil, Turkey, India and China for the period 1999 to
2009 from DataStream. The sample firms meet standard criteria employed widely in the
literature: stock prices are available for December of year t-1 and June of year t, and book
7 The year of equity market liberalization in the BTIC is as follows: Brazil, 1991; China, 1995; India, 1992;
and Turkey, 1989 (Kim and Kenny, 2007).
11
value for year t-1; and each firm has at least two years’ complete data.8 Table A1 in the
Appendix shows the number of firms belonging to each portfolio by country and year.
In order to identify the existence of a value premium, we employ the standard Fama
and French (1993) methodology. We form six portfolios (S/L, S/M and S/H; B/L, B/M and
B/H) by intersecting two groups that are arranged by ME = market value of equity, and
BE/ME where BE = net tangible assets (equity capital plus reserves minus intangibles).
Stocks with ME higher than the median are classified as ‘big’ (B); those with ME below the
median are classified ‘small’ (S). For BE/ME the three breakpoints are bottom 30% (‘Low’),
middle 40% (‘Medium’) and top 30% (‘High’).9 We calculate value-weighted monthly
returns for the six portfolios from July of year t to June of year t+1, when we re-form the
portfolios.
In order to explore the sources of the value premium, we divide our analysis into two
stages. In the first stage we use measures of bankruptcy risk, proposed by Altman (1993), to
investigate whether firms with a high likelihood of distress, measured by the Z-score, also
have a high BE/ME or value premium.10
Fama and French (1993) and Chen and Zhang
(1998) offer empirical evidence to suggest high value firms are assigned a higher risk
premium because they have a higher probability of distress, implying that bankruptcy risk is a
systemic factor. Dichev (1998) and Griffin and Lemmon (2002) offer empirical evidence to
reject the conjecture that bankruptcy risk is systematic and rewarded by higher returns. In
8 This criterion is required to address the issue of survival bias (see Banz and Breen, 1986; Kothari, et al.,
1995).
9 We do not use negative book equity (BE) firms when forming the size-BE/ME portfolios for lack of
economic explanation.
10 We employ the Altman’s (1993) model to evaluate the Z-score:
Z = 1.2X1 + 1.4 X2 + 3.3X3 + 0.6X4 + 1.0X5
X1 = Working Capital / Total Assets
X2 = Retained Earnings / Total Assets
X3 = Earnings Before Taxes + Interest / Total Assets
X4 = Market value of equity / Total Liabilities
X5 = Net Sales / Total Assets
12
what follows, an inverse relationship between the Z-score and BE/ME ratio suggests the
variables capture information related to a priced distress factor as in Fama and French (1993)
and Chen and Zhang (1998). Alternatively, a positive relationship implies the two variables
contain different information that is potentially related to differences in relative risk across
firms (see Dichev, 1998; Griffin and Lemmon, 2002). In the context of this paper, we argue
that the relative risk results from the idiosyncratic characteristics of each firm.
For the analysis of bankruptcy risk and value, we follow Griffin and Lemmon (2002)
and construct a set of portfolios that we form according to the three indicators of BE/ME
(small, medium and high value), the five quintiles of bankruptcy risk measured by the Z-
score, and two size (ME) measures. The breakpoints for BE/ME are the 30th
and 70th
percentile points. For reporting purposes we show size-adjusted data, which are the simple
averages of the means of the small and large stocks. Firms located in the lowest Z-score
quintile have the highest probability of bankruptcy.
In the second stage we use 36-month rolling regressions to estimate the Fama and
French (1995) three factor model to derive time varying HML coefficients for 25 portfolios
sorted on size and BE/ME.11
Changes in the loading over time reflect changes in business
and financial risk. Subsequent analysis uses the estimated HML coefficients on growth and
value portfolios as dependent variables. HML is interpreted to proxy firm characteristics
which implies common variation in returns arises because individual portfolios comprise
similar stocks with similar factor loadings irrespective of whether a firm is distressed or not.
This interpretation follows Daniel and Titman (1997) and it contrasts the view of Fama and
French (1993) that HML is proxy for distress probability. A preliminary analysis shows the
Fama and French (1995) model provides an adequate description of portfolio returns for each
11 The three factor model is given by: Rpt-Rft = α + b (Rmt-Rft) + s SMB + h HML + ε.
13
of the BTIC since none of the estimated alpha coefficients are significantly different from
zero (see Table 1).12
[Insert Table 1 here]
We use panel data models to estimate the time-varying coefficient on HML for the i’th
portfolio at time t on a set of conditioning variables known at time t. The conditioning
variables are the current and lagged values of the natural logarithms of total assets and total
debt. The assets variable captures the sensitivity of the coefficient on HML to growth whilst
the debt variable measures sensitivity to leverage. GDP growth controls for macroeconomic
conditions. Interaction variables allow for differences in the coefficients for value and
growth portfolios. A Hausman test selects between fixed-effects and random-effects models.
V. Empirical Results
Table 2 reports the weighted monthly excess returns of the six size-BE/ME portfolios
constructed using the three factor model (Fama and French, 1993) for each of the BTIC. For
portfolios constructed using “big” stocks, we find statistically significant excess returns to
value firms in excess of returns to growth firms in Turkey, India and China. In Brazil, excess
return is highest for the intermediate band. For portfolios constructed using “small” stocks,
excess returns are significant and higher for value firms in Brazil and China. Although this
pattern repeats in Turkey the returns are not significant. In contrast, returns are significant
and highest for growth firms in India. A positive and significant VMG (value minus growth)
confirms the findings for Brazil, Turkey and China. VMG measures the difference in excess
returns between value and growth portfolios and equals (B/H+S/H)/2–(B/L+S/L)/2. Our
12 The average value of beta is different from one depending on the orientation of firms in the economy. If
the aggregate firm in the economy is value oriented then beta will be less than 1. This is consistent with
studies which demonstrate that stocks with below market risk (low beta) yield higher risk adjusted return
than predicted by the theoretical CAPM (see Black et al., 1972; Miller and Scholes, 1972), and evidence
showing value firms have lower beta than growth firms (see Capaul et al., 1993).
14
evidence infers the existence of a value premium in each country except India. We
conjecture the fast-growing information technology industry in India, containing many small
high-technology firms, might account for the exceptionally high excess return in the SG
portfolio.
[Insert Table 2 here]
In Table 3 we construct the portfolios as follows. Portfolios are sorted into small and
large categories, and interlocked with value (using breakpoints at the 30th
and 70th
percentile
points to define low, medium and high value) and the probability of bankruptcy as indicated
in quintiles of Z-scores. Table 3 reports size-adjusted data, the simple average of small and
large stocks, for bankruptcy risk, size, returns, and leverage.
Panel A shows little variation in the mean Z-scores between low value and high value
portfolios in each country except for Indian firms in quintile one for low value stocks (-0.262)
relative to their high value counterparts (0.803). The lowest (highest) Z-scores are in Brazil
(Turkey). In the higher quintiles and across countries, low and high values stocks achieve
similar Z-scores, which suggest the presence of value premium is not related to distress risk
thereby contradicting Fama and French (1995). Panel B shows the average book-to-market
ratio of the three stock groupings by quintiles of Z-score. On average in Brazil, India and
China, high value firms in the lowest quintile achieve considerably larger book-to-market
ratios, which again is inconsistent with the Fama and French hypothesis.
[Insert Table 3 here]
Table 3 also reports statistics on profitability, leverage, size (capitalization), and total
assets for firms in each portfolio in order to further examine the hypothesis that the Z-score
and BE/ME are related to characteristics purported to reflect distress risk. Panel C shows
mean profitability is positively related to the Z-score, and is larger for high value firms in
15
each country (except Turkey). Leverage is inversely related to the Z-score in each country
(see Panel D). The data show high value and high risk firms are more heavily levered
particularly in Brazil and India, which supports our earlier argument that value firms use
more debt financing than growth firms. In Brazil and Turkey, low value stocks tend to be
better capitalized (see Panel E) and larger in terms of total assets whereas high value firms
are larger in terms of assets size in China and India (see Panel F).
Table 4 shows the results of rolling regressions estimations of the Fama and French
three factor model for the six portfolios. Since our focus is on the coefficient of HML (h), we
do not discuss other coefficients. HML is defined as the difference between the simple
average returns on two high value portfolios (S/H and B/H) and two low value portfolios (S/L
and B/L). All B/M portfolios (except China), B/H and S/H portfolios are associated with
positive values on HML. That we observe positive loadings on value-orientated portfolios in
each country demonstrates the presence of the value premium in those markets. We observe
negative coefficients on HML for low value portfolios B/L and S/L across countries, and also
for portfolio S/M (except Turkey). The findings demonstrate a difference in returns between
portfolios of growth stocks ranked by firm size.
[Insert Table 4 here]
In spite of the apparent heterogeneity in loadings, our findings - at least for Brazil and
China - are consistent with the contention of Fama and French (1995) that growth [value]
stocks should have negative [positive] loadings. In contrast to Fama and French (1995) but
consistent with Daniel and Titman (1997), we suggest the negative loading on growth firms is
not a function of distress as our earlier analysis indicates the level of distress is comparable
for growth and value firms. Rather, we that contend growth portfolios have lower loadings
because the choice of delaying growth options gives growth firms the opportunity to reduce
their risk. In addition, delaying the exercise of these options enables growth firms to
16
accumulate and/or hoard cash. Our results offer support for claims that growth portfolios
earn low returns because of a cash drag effect since cash generates very little return.
We contend that investors display an ‘infatuation’ with growth firms based on the
potential growth opportunities stemming from embedded growth options. Thus, stock prices
are driven upwards through bidding which contrasts with the ‘unspectacular’ value firms. In
Lucas’ Rational Expectations framework (1978), growth firms constitute an ‘alluring’ asset, a
point further extended by Sargent (1987). This interpretation reconciles the neoclassical and
behavioral perspectives. The leverage of value firms causes a drag on their performance in
poor economic environments particularly as leveraged equity displays volatility associated
with financial options (see Merton, 1974). The expansion of the economy creates a bounce-
back effect on value stocks, helping to reconcile the neoclassical perspective with the
behavioral as espoused by DeBondt and Thaler (1985, 1987).
In light of our intuition, we graph the coefficients of HML for twenty of the 25
portfolios to provide further insight to our argument.13
Figure 1a-d illustrate the patterns of
time varying betas (coefficients of HML) for growth and value portfolios. The figures, with
the exception of Turkey, demonstrate that value portfolios have higher beta and are more
stable over time compared to growth portfolios.
[Insert Figure 1a-d here]
The exercise we report on above uses rolling regressions to estimate the three factor
model. To check the robustness of the results, we re-estimate the models using static panel
data techniques. A Hausman test finds the χ2 statistic equal to 38.48 and statistically
significant at the 1 percent level. The test result shows the fixed effects model is preferred to
its random effects counterpart. Table 5 reports the fixed-effects estimation.
13 We categorize portfolios labelled with L1 and L2 as growth portfolios and L4 and L5 as value portfolios.
Portfolios labelled with L3 are not graphed because they are neither growth nor value. In Figure 1a-d we
present four graphs per country rather than 25. The remainder are available from the authors upon request.
17
Ex ante we expect growth portfolios to exhibit sensitivity to size (total assets) whereas
value portfolios should display sensitivity to leverage. Using a binary variable to identify
value portfolios, we create interaction variables between this indicator and size and leverage,
at contemporaneous and lagged values for two periods to determine separate value portfolio
effects. Lastly, the natural logarithm of GDP growth controls for business cycle effects.
The results show a positive and significant relation between HML and total assets
lagged two periods at the 10 per cent significance level. However, the estimated coefficients
on the interaction variables with total assets (6 and 8) are negatively signed and statistically
significant (except 7 the coefficient on the one period lag interaction). The finding confirms
the differential between value and growth portfolios in terms of the sensitivity of HML with
respect to size. The estimated coefficient on leverage is negative (though insignificant) and
suggests more highly levered portfolios are less sensitive. However, the interaction term
between contemporaneous leverage and HML (9) is positive and highly significant, which
signals value portfolios are very sensitive to leverage.
[Insert Table 5 here]
VI Robustness
We use an alternative procedure to construct portfolios to test the robustness of the
results and construct twenty five intersecting size-value portfolios using the quintile
breakpoints for ME and BE/ME and equally weighted monthly returns instead of value-
weighted returns. To illustrate, using ‘S’ to indicate size and ‘L’ to indicate BE/ME, the
portfolio S1L1 contains stocks that are ranked in the first quartile (less than 20%) of both size
and value. Table A2 in the Appendix presents the data for the 25 portfolios and confirms the
finding in Table 1 which demonstrates the Fama and French (1995) model adequately
describes the portfolio returns. Table A3 reports the simple excess returns on the 25
18
portfolios by country and checks the robustness of findings reported in Table 2. The
distribution of excess returns across portfolios lets us formulate some generalizations: first,
excess returns on the highest value firms tend to be greater in comparison with low value
firms; second, the distinction in returns by size is less marked and varies across countries.
Table A3 shows the smallest firms achieve significant returns, which for some portfolios,
exceed returns on larger firms. Both sets of our results confirm findings from several
previous studies that show high value stocks achieve larger mean expected returns. In Table
A4 we report robustness checks for findings in Table 4 using the set of 25 interlocking
portfolios. The estimated HML coefficients exhibit intra and inter heterogeneity across the
BTIC although some generalized country specific patterns are apparent. For instance, the
value premium is supported by the positive loadings for medium-to-large sized and medium-
to-high value portfolios in Brazil (except for the portfolio containing the largest and highest
value stocks). In the case of China, we observe positive loadings for a limited number of
portfolios, namely, portfolios of mid-to-large sized firms in the upper two quartiles of value.
The loadings on portfolios in India and Turkey are all positive (with two exceptions for
Turkey) yet it is difficult to discern hard patterns across size and value quintiles. Lastly, we
observe the magnitude of loadings for India and Turkey to be greater relative to Brazil and
China.
VII Conclusion
A number of theories are postulated to rationalize the source of the value premium yet
the issue remains controversial. We reassess this issue for the BTIC group of countries that
are characterized as having vast economic potential. The paper rationalizes the value
premium in terms of economic fundamentals, attributing the premium to the investment
patterns of growth firms that we contend are more likely to hoard cash, particularly during
19
episodes of economic malaise. Although this behaviour is understood because it limits
growth firms’ exposure to risk, nevertheless it negatively impacts on both their market
valuation and returns.
The paper helps to reconcile the diverging neoclassical views of Fama and French
(1995) and Daniel and Titman (1997) in explaining the expected returns of value and growth
stocks. Fama and French (1995) claim the HML coefficient measures distress risk whereas
Daniel and Titman (1997) believe it captures firm characteristics. Our evidence offers
support for the latter interpretation. We contest that growth firms are endowed with growth
options, which entails capital outlay whilst enhancing business risk and it is this feature that
differentiates growth firms from value firms. Value firms, by contrast, use fixed assets as
collateral to lever up in order to boost earnings, which in turn aggravates financial risk. This
interpretation of our findings is consistent with Chen et al. (2011).
We also reconcile the diverging neoclassical and behavioral perspectives by invoking a
rational expectations perspective (Lucas, 1978) as extended by Sargent (1987). We consider
the range of options available to growth firms provides a utility (‘infatuation’) that is separate
from monetary returns in the forms of capital gains and dividends. This inherent utility of
growth firms is attractive to investors and causes their stock prices to appreciate, which
subsequently lowers returns.
Our empirical evidence offers three conclusions. Our first result re-confirms the
presence of the value premium in emerging markets under favorable economic conditions.
Second, value stocks and growth stocks are not characterized by different levels of distress as
suggested by Fama and French (1995). We observe value firms are more highly levered than
growth firms, which reconciles the behavioral perspective (DeBondt and Thaler, 1985, 1987)
with the neoclassical perspective (Merton, 1974). We contend the leverage behavior of value
firms exhibits characteristics similar to volatile financial options, which plummet very fast
20
during economic downturns and rebound equally fast upon recovery. Our findings are robust
to alternative means of portfolio construction.
Third, by observing the time varying pattern of conditional beta (HML) we find value
(growth) portfolios are less (more) sensitive to size but more (less) sensitive to leverage. As
a robustness check, fixed-effects methods demonstrate the value premium is attributable to
economic fundamentals in a static framework. The finding that growth portfolios are
sensitive to changes in total assets reaffirms our belief that the risk and return structure of
growth firms is determined by their investment pattern. We believe our paper provides
further insights on the source of the value premium, particularly in the context of the under-
researched emerging economies.
21
References
Altman, E.I., 1993. Corporate financial distress and bankruptcy. John Wiley and Sons, Inc.,
New York.
Banz, R., 1981. The relationship between return and market value of common stocks. Journal
of Financial Economics 9, 3-18.
____, Breen, W.J., 1986. Sample-dependent results using accounting and market data: Some
evidence. Journal of Finance 41, 779-93.
Barber B., Lyon J.D., 1997. Firm size, book-to-market ratio, and security returns: A holdout
sample of financial firms. Journal of Finance 52, 875-884.
Barry C.B., Goldreyer E., Lockwood L., Rodriguez M., 2002. Robustness of size and value
effects in emerging equity markets, 1985-2000. Emerging Markets Review 3, 1-30.
Basu, S., 1977. Investment performance of common stocks in relation to their price-earnings
ratio: A test of the efficient market hypothesis. Journal of Finance 32, 663-682.
____, 1983. The relationship between earnings yield, market value, and return for NYSE
common stocks: Further evidence. Journal of Financial Economics 12, 129-156.
Bekaert, G., Harvey, C.R., 2003. Emerging markets finance. Journal of Empirical Finance 10,
3-55.
Bekaert, G., Harvey, C.R., Lundblad, C.T., 2003a. Equity market liberalization in emerging
markets. Federal Reserve Bank of St Louis Review 85, 4, 53-74.
Bekaert, G., Harvey, C.R., Lumsdaine, R.L., 2003b. Dating the integration of world equity
markets. Journal of Financial Economics 65, 203-247.
Bhandari, L., 1988. Debt/Equity ratio and expected common stock returns: Empirical
evidence. Journal of Finance 43, 507-528.
Black, F., 1993. Beta and return. Journal of Portfolio Management 20, 8-18.
22
____, Jensen, M.C., Scholes, M., 1972. The capital asset pricing model: Some empirical tests.
In: Jensen, M.C. (ed.) Studies in the theory of capital markets, Praeger Publishers, New
York, USA, 79-121.
Blume, M., Friend, I., 1973. A new look at the capital asset pricing model. Journal of Finance
28, 19-33.
Campbell J.Y., 2000. Asset pricing at the millennium. Journal of Finance 55, 1515-1567.
Capaul, C., Rowley, I., Sharpe, W.F., 1993. International Value and Growth Stock Returns.
Financial Analysts Journal 49, 27-36.
Chan, L.K., Jegadeesh, N., Lakonishok, J., 1995. Evaluating the performance of value versus
glamour stocks: The impact of selection bias. Journal of Financial Economics 38, 269-296.
Chen L., Novy-Marx R., Zhang L., 2011. An alternative three-factor model. Available at
SSRN: http://ssrn.com/abstract=1418117 or http://dx.doi.org/10.2139/ssrn.1418117.
Chen, N., Zhang, F., 1998. Risk and return of value stocks. Journal of Business 71, 501-535.
Chou, P-H., Ko, K-C., Lin, S-J, 2011. Factors, Characteristics and Endogenous Structural
Breaks: Evidence from Japan. Available at SSRN: http://ssrn.com/abstract=1911691 or
http://dx.doi.org/10.2139/ssrn.1911691.
Cohen, R.B., Gompers, P.A., Vuolteenaho, T., 2002. Who under reacts to cash-flow news?
Evidence from trading between individuals and institutions. Journal of Financial Economics
66, 409-462.
____, Polk, C.K., 1995. COMPUSTAT selection bias in tests of the Sharpe-Lintner-Black
CPAM. Working paper, University of Chicago.
Conrad, J., Cooper, M., Kaul, G., 2003. Value versus glamour. Journal of Finance 58, 1969-
1996.
Davis, J.L., 1994. The cross-section of realized stock returns: The pre-COMPUSTAT
evidence. Journal of Finance 50, 1579-1593.
23
Daniel K., Titman S., 1997. Evidence on the characteristics of cross sectional variation in
stock returns. Journal of Finance 52, 1-33.
De Bondt, W.R.M., Thaler, R.H., 1985. Does the stock market overreact? Journal of Finance
40, 793-805.
____________, 1987. Further evidence on investor overreaction and stock market
seasonality. Journal of Finance 42, 557-581.
de Groot C.G.M., Verschoor, W.F.C., 2002. Further evidence on Asian stock return
behaviour. Emerging Markets Review 3, 179-193.
Dichev, I.D., 1998. Is the risk of bankruptcy is a systematic risk? Journal of Finance 53, 3,
1131-1147.
Ding, D.K., Chua. J.L., Fetherston T.A., 2005. The performance of value and growth
portfolios in East Asia before the Asian financial crisis. Pacific-Basin Finance Journal 13,
185-199.
Economist, 2011. Economic Focus: Why the tail wags the dog. The Economist (6th
August
2011), 63.
Fama, E.F., French, K.R., 1992. The cross-section of expected returns. Journal of Finance 47,
427-465.
___________, 1993. Common risk factors in the returns on stocks and bonds. Journal of
Financial Economics 33, 3-56.
___________, 1995. Size and book-to-market factors in earnings and returns. Journal of
Finance 50, 131-156.
___________, 1998. Value versus growth: the international evidence. Journal of Finance 53,
1975-1999.
____, MacBeth, J.D., 1973. Risk, return, and equilibrium: Empirical test. Journal of Political
Economy 81, 607-637.
24
Griffin, J.M., Lemmon, M.L., 2002. Book-to-Market, Equity, Distress Risk, and Stock
Returns. Journal of Finance 57, 2317-2336.
Gulen, H., Xing, Y., Zhang, L., 2011. Value versus Growth: Time-Varying Expected Stock
Returns. Financial Management Summer 2011, 381–407.
Guo, H., Savickas, R., Wang, Z., Yang, J., 2009. Is the Value Premium a Proxy for Time-
Varying Investment Opportunities? Some Time-Series Evidence. Journal of Financial and
Quantitative Analysis 44(1), 133–154.
Haugen, R., 1995. The race between value and growth. In: The new finance: The case against
efficient markets. Prentice-Hall, New York, 55-71.
Hughes, J., 2011. Bric’ creator adds newcomers to list. Financial Times (January 11, 2011),
Available at:<http://www.ft.com > [Accessed 20 May 2011]
Ingersoll, Jr. J.E., Ross, S.A., 1992. Waiting to invest: investment and uncertainty. Journal of
Business 65, 1-29.
Jegadeesh, N., Titman, S., 1993. Returns to buying winners and selling losers: Implications
for market efficiency. Journal of Finance 48, 65-91.
Kim, B., Kenny, L.W., 2007. Explaining when developing countries liberalize their financial
equity markets. Journal of International Financial Markets, Institutions and Money 17, 387-
402.
Kothari, S., Shanken, J., Sloan, R., 1995. Another look at the cross-section of expected stock
returns. Journal of Finance 50, 185-224.
Lakonishok, J., Shleifer, A., Vishny, R.W., 1994. Contrarian investment, extrapolation, and
risk. Journal of Finance 49, 1541-1578.
La Porta, R., Lakonishok, J., Shleifer, A., Vishny, R.W., 1997. Good news for value stocks:
Further evidence on market efficiency. Journal of Finance 52, 859-874.
25
Lee, E., Liu, W., Strong, N., 2007. UK evidence on the characteristics versus covariance
debate. European Financial Management 13, 742-756.
Levine, R., Zervos, S., 1998. Stock markets, banks and economic growth. American
Economic Review 88, 3, 537-558.
Li, X., Brooks, C., Miffre, J., 2009. The Value Premium and Time-Varying Volatility.
Journal of Business Finance & Accounting 36(9) & (10), 1252–1272.
Lintner, J., 1965a. Security prices, risk and maximal gains from diversification. Journal of
Finance 20, 587-615.
____, 1965b. The valuation of risky assets and the selection of risky investments in stock
portfolios and capital budgets. Review of Economics and Statistics 47, 13-37.
Lo, A., MacKinlay, A.C., 1988. Stock market prices do not follow random walks: Evidence
from a simple specification test. Review of Financial Studies 1, 41-66.
Lucas, R.E. Jr., 1978. Asset prices in an exchange economy. Econometrica 46, 1426-1445.
MacKinlay, A.C., 1995. Multifactor models do not explain deviations from the CAPM.
Journal of Financial Economics 38, 3-28.
McDonald, R., Siegel, D., 1986. The value of waiting to invest. The Quarterly Journal of
Economics 101, 707-728.
Merton, R.C., 1973. An intertemporal capital asset pricing model. Econometrica 41, 129-176.
____, 1974. On the pricing of debt: The risk structure of interest rate. Journal of Finance 29,
449-470.
Miller, M.H., Scholes, M., 1972. Rates of Return in Relation to Risk: A Re-examination of
some Recent Findings. In: Jensen, M.C. (ed.) Studies in the theory of capital markets,
Praeger Publishers, New York, USA, 60-62.
Mossin, J. 1966. Equilibrium in a Capital Asset Market. Econometrica 34, 768-783.
26
Roll, R., 1977. A critique of the asset pricing theory's tests, Part I: On past and potential
testability of the theory. Journal of Financial Economics 4, 129-176.
Rosenberg, B., Reid, K., Lanstein. R., 1985. Persuasive evidence of market inefficiency.
Journal of Portfolio Management 11, 9-17.
Ross, S.A., 1976. The arbitrage theory of capital asset pricing. Journal of Economic Theory
3, 343-362.
Rouwenhorst, G.K., 1999. Local return factors and turnover in emerging stock markets.
Journal of Finance 54, 1439-1464.
Sargent, T.J. 1987. Dynamic Macroeconomic Theory. Harvard University Press, Cambridge,
MA, USA, 123.
Sharpe, W.F., 1964. Capital asset prices: A theory of market equilibrium under conditions of
risk. Journal of Finance, 19, 425-442.
Stattman, D., 1980. Book values and stock returns. The Chicago MBA: A Journal of Selected
Papers 4, 25-45.
Stivers, C., Sun, L., 2010. Cross-Sectional Return Dispersion and Time Variation in Value
and Momentum Premiums. Journal of Financial and Quantitative Analysis, 45(4), 987–
1014.
Titman, S., 1985. Urban land prices under uncertainty. The American Economic Review 75,
505-514.
____, Wei, K. C. J., Xie, F., 2004. Capital investments and stock returns. Journal of Financial
and Quantitative Analysis 39, 677-700.
World Development Indicators (2010). [Accessed 28 June 2011]
Yen, J.Y., Sun, Q., Yan Y., 2004. Value versus growth stocks in Singapore. Journal of
Multinational Financial Management 14, 19-34.
Zhang, L., 2005. The Value Premium. Journal of Finance 60, 67-103.
27
Table 1: Weighted Monthly Excess Returns for Portfolios Sorted on Size (ME) and
Value (BE/ME) - July 1999 to June 2009, 120 Months (%)
Stocks with ME above the median are deemed ‘big’ whilst stocks with ME below the median are
‘small’. The breakpoints for BE/ME are the 30th and 70
th percentiles. The intersection produces six
portfolios: B/L = big-low value; B/M = big-middle value; B/H = big-high value; S/L = small-low
value; S/M = small-middle value, S/H = small-high value. Returns are generated using the Fama and
French (1993) three factor model: Rpt - Rft = α + b (Rmt-Rft) + s SMB + h HML + ε.
*,**,*** indicate statistical significance at the 10, 5, and 1 percentage levels.
Portfolio B/L B/M B/H S/L S/M S/H
Brazil
α 1.89*** 0.52 1.69*** 1.16 1.35* 1.89***
b 0.22 0.86*** 0.04 0.13 0.21* 0.24
s -0.28** -0.68*** -0.70*** 0.86*** 0.62** 0.12
h -0.10 0.54*** 0.56*** -0.73*** -0.26 0.82***
Turkey
α 0.52*** -2.18*** -0.04 -0.06 -1.18*** -0.31
b 1.26*** 0.09*** 0.35*** 1.06*** 0.54*** 0.63***
s -0.30*** 0.98*** 0.58*** 1.88*** 1.10*** 1.25***
h -0.32*** 0.92*** 0.63*** -0.07 0.38*** 0.39***
India
α -0.30* 0.63*** 0.92 1.64** 0.05 -0.08
b 0.86*** 1.12*** 1.01*** 1.34*** 0.83*** 0.98***
s -0.04* -0.07*** 0.10 2.79*** 0.44*** 0.74***
h -0.02 0.01 0.13* -1.94*** 0.21** 0.48***
China
α -0.17 -0.02 -0.01 -0.00 0.12 -0.19**
b 0.96*** 1.03*** 1.01*** 0.98*** 0.96*** 0.98***
s 0.02 -0.22*** -0.09 0.87*** 0.78*** 0.96***
h -0.48*** -0.21*** 0.57*** -0.39*** 0.00 0.43***
28
Table 2: Weighted Monthly Excess Returns on Portfolios Sorted on Size and Value
– July 1999 to June 2009, 120 Months: Summary Statistics (%)
Stocks with ME above the median are deemed ‘big’ whilst stocks with ME below the median are
‘small’. The breakpoints for BE/ME are the 30th and 70
th percentiles. The intersection produces six
portfolios: B/L = big-low value; B/M = big-middle value; B/H = big-high value; S/L = small-low
value; S/M = small-middle value, S/H = small-high value. VMG = value minus growth and is given
by [{(S/H+B/H)/2}-{(S/L+B/L)/2}]. RPTRFT = return on portfolio minus the risk free rate. Returns
are generated using the Fama and French (1993) three factor model: Rpt - Rft = α + b (Rmt-Rft) + s SMB
+ h HML + ε.
We test Ho = 0; H1 > 0. *,**,*** indicate significance at the 10, 5, and 1 percentage levels.
Portfolio RPTRFT Portfolio RPTRFT
Brazil
B/L 2.13*** S/L 1.38**
B/M 2.21*** S/M 2.10***
B/H 1.82*** S/H 3.48***
VMG 0.895*
Turkey
TURKEY
Turkey B/L -0.84 S/L 1.19
B/M 0.62 S/M 0.47
B/H 1.70** S/H 1.48
VMG 1.415*
India
B/L 0.81 S/L 8.01**
B/M 2.10** S/M 2.67***
B/H 2.81*** S/H 3.91***
VMG -1.05
China
B/L 0.12 S/L 0.76
B/M 0.37 S/M 1.09*
B/H 0.92* S/H 1.13*
VMG 0.585***
29
Table 3: Firm Characteristics for Portfolios Sorted on Value, Distress Probability and
Size - Summary Statistics
Portfolios are formed by intersecting BE/ME (low, medium, high - the breakpoints are at the 30th and
70th percentile points); five quintiles of bankruptcy risk (Z-score); and two size (ME) measures (above
and below the median). We report size-adjusted data i.e. the simple averages of the means of the small
and large stocks. Firms from July 1999 to June 2009 are ranked independently every June.
BE/ME Low Medium High Low Medium High
Panel A: Z-score
Z-score Brazil Turkey
1 0.560 0.572 0.418 0.735 0.811 0.772
2 1.414 1.389 1.608 2.105 2.160 2.051
3 2.060 2.303 2.019 3.517 3.486 3.524
4
5
3.590
7.201
3.157
8.710
3.040
8.162
8.939
69.948
8.810
62.708
7.979
69.621
India China
1 -0.262 0.940 0.803 1.147 1.454 1.352
2 1.811 1.898 1.738 2.757 2.645 2.575
3 2.910 2.911 2.490 4.277 4.145 3.891
4
5
5.320
20.919
4.312
18.018
5.190
17.388
7.185
15.893
6.760
16.027
6.689
16.800
Panel B: BE/ME (book-to-market equity)
Z-score Brazil Turkey
1 0.563 2.236 13.40 0.424 1.071 2.676
2 0.425 2.074 10.04 0.439 1.079 2.265
3 0.528 2.023 8.526 0.509 1.059 1.986
4
5
0.467
0.542
2.060
1.917
9.570
9.191
0.429
0.431
1.046
1.021
2.401
2.110
India China
1 0.233 0.915 12.935 0.169 0.391 2.029
2 0.241 0.826 5.477 0.186 0.386 0.890
3 0.282 0.769 9.165 0.189 0.379 0.784
4
5
0.242
0.211
0.807
0.788
2.640
1.800
0.185
0.194
0.376
0.382
0.730
0.664
Panel C: RoA (income before extraordinary items and tax/total assets)
Z-score Brazil Turkey
1 -0.561 2.582 4.204 2.129 2.518 1.161
2 8.242 7.777 10.039 10.329 11.724 8.022
3 11.840 11.093 9.718 10.876 8.793 10.851
4
5
12.801
12.895
11.389
12.508
12.408
16.960
11.407
11.201
11.779
11.186
11.443
10.703
India China
1 1.279 1.113 4.326 -3.133 1.824 2.796
2 3.016 7.057 8.916 1.942 3.603 4.181
3 8.164 9.622 10.18 3.184 4.505 4.971
4
5
11.293
10.369
13.055
13.715
16.138
5.989
4.574
5.783
5.539
5.913
5.836
5.427
30
BE/ME Low Medium High Low Medium High
Panel D: Leverage (total assets - book equity / market value of equity)
Z-score Brazil Turkey
1 2.373 5.260 13.883 38.514 47.914 17.290
2 1.687 3.526 12.184 2.407 3.945 1.839
3 2.347 2.235 4.987 1.595 1.887 3.097
4
5
0.664
0.294
1.401
0.311
2.871
0.686
0.192
0.019
0.195
0.0228
0.542
0.017
India China
1 1.851 3.648 22.492 0.683 0.897 2.261
2 0.773 1.412 3.951 0.375 0.458 0.736
3 0.566 0.671 11.65 0.249 0.299 0.366
4
5
0.210
0.084
0.513
0.111
0.482
0.054
0.142
0.604
0.171
0.060
0.187
0.067
Panel E: Size (US$ m, market value of equity)
Z-score Brazil Turkey
1 1,159 785 126 443 116 47
2 1,515 1,629 572 447 194 141
3 3,491 1,594 491 664 381 205
4
5
10,800
2,558
5,086
642
166
71
1,921
1,575
273
698
116
571
India China
1 11,000 15,400 3,463 2,969 4,116 3,682
2 46,800 9,978 3,201 3,106 4,171 5,050
3 27,100 30,100 4,202 3,518 6,638 6,306
4
5
81,600
33,000
95,200
8,027
3,381
692
4,495
4,147
8,669
12,100
5,073
3,314
Panel F: Total Assets (US$ m, book value)
Z-score Brazil Turkey
1 3,339 3,995 1,800 6,824 3,845 3,232
2 3,776 7,321 3,066 1,135 1,600 938
3 4,274 4,914 1,827 1,215 1,201 871
4
5
14,600
2,212
8,870
799
975
245
1,099
389
433
323
265
243
India China
1 3,9200 49,700 53,200 3,292 6,413 8,261
2 5,6100 27,600 23,400 2,265 4,077 7,660
3 2,0200 32,222 56,500 2,058 5,076 8,662
4
5
7,0700
1,0300
55,000
8,255
10,200
2,416
1,946
1,549
6,623
4,345
5,983
3,035
31
Table 4: Mean Coefficients of Rolling Regressions: Portfolios Sorted on Size and Value
Stocks with ME above the median are deemed ‘big’ whilst stocks with ME below the median are
‘small’. The breakpoints for BE/ME are the 30th and 70
th percentiles. The intersection produces six
portfolios: B/L = big-low value; B/M = big-middle value; B/H = big-high value; S/L = small-low
value; S/M = small-middle value, S/H = small-high value. The coefficients are estimated from 36
month rolling regressions of the Fama and French (1993) three factor model: Rpt - Rft = α + b (Rmt-Rft)
+ s SMB + h HML + ε. Rm - Rf is the return on the market portfolio less the risk free rate. SMB is the
difference between the simple average of returns on small stock portfolios (S/L, S/M, S/H) and big
stock portfolios (B/L, B/M, B/H). HML is the difference between the simple average of returns on
two high value portfolios (S/H, B/H) and returns on two low value portfolios (S/L, B/L).
B/L B/M B/H S/L S/M S/H
Brazil
Constant, α 1.497 0.869 1.135 0.611 1.062 1.408
Rm-Rf, b 0.566 0.697 0.201 0.158 0.232 0.582
SMB, s -0.046 -0.828 -0.900 1.115 0.324 0.412
HML, h -0.199 0.599 0.782 -0.950 -0.026 0.666
Turkey
Constant, α 0.544 -2.233 0.126 -0.093 -1.108 -0.457
Rm-Rf, b 1.210 0.037 0.730 0.885 0.596 0.568
SMB, s -0.308 0.996 0.061 1.981 0.912 1.265
HML, h -0.296 0.977 0.283 0.090 0.335 0.438
India
Constant, α -0.221 0.487 1.124 1.294 0.026 -0.050
Rm-Rf, b 0.918 1.067 1.006 1.194 0.987 0.968
SMB, s 0.141 -0.245 0.006 2.052 1.134 0.645
HML, h -0.243 0.200 0.373 -1.416 -0.342 0.532
China
Constant, α 0.001 0.043 -0.015 -0.048 -0.217 -0.087
Rm-Rf, b 1.002 1.008 1.009 0.983 0.907 0.986
SMB, s -0.185 -0.199 -0.178 0.749 0.518 0.892
HML, h -0.639 -0.223 0.538 -0.422 -0.064 0.337
32
Table 5: Fixed effects regression of the sensitivity of HML
HML is the difference between the simple average of returns on two high value portfolios (S/H, B/H)
and returns on two low value portfolios (S/L, B/L). It is calculated for 20 portfolios sorted on size
(BE) and value (BE/ME) using quintile breakpoints, and is recalculated when the portfolios are
rebalanced. We use a fixed effects regression to estimate the model:
HML = α + β0 Total Assetsit + β1 Total Assetsit-1 + β2 Total Assetsit-2 + β3Leverageit + β4Leverageit-1 +
β5Leverageit-2+ β6Total Assets * Dit + β7Total Assets * Dit-1 + β8Total Assets * Dit-2 + β9Leverage * Dit
+ β10Leverage * Dit-1 + β11Leverage * Dit-2 + β12 GDPt + ηi + ηt + εit
Where: ηi is an unobserved portfolio-specific effect and ηt captures common period-specific effects; εit
is an error term, which represents measurement errors and other explanatory variables that have been
omitted. It is assumed to be independently identical normally distributed with zero mean and constant
variance. D is a dummy variable taking the value of unity if the portfolio is value and 0 otherwise.
*, **, *** indicate significance at 10%, 5% and 1% respectively.
Independent Variables Coefficient t
Total Assets 0.0442 0.50
Total Assetst-1
Total Assetst-2
-0.0403
0.1651*
-0.41
1.80
Total Leverage -0.0910 -1.48
Total Leveraget-1
Total Leveraget-2
0.0613
-0.0797
0.81
-1.19
Total Assets * D -0.2980** -2.06
Total Assets * Dt-1 -0.1534 -0.97
Total Assets * Dt-2 -0.2982** -2.04
Total Debt * D 0.2932*** 2.67
Total Debt * Dt-1 0.0748 0.62
Total Debt * Dt-2
GDP growth
0.1572
-0.0034
1.43
-0.41
Constant 1.0999 1.45
33
Figure 1a-d
1a. Patterns of HML for Portfolios sorted on Size and BE/ME: Brazil
1b. Patterns of HML for Portfolios sorted on Size and BE/ME: Turkey
-6-4
-20
24
S1
L1
2001m1 2003m1 2005m1 2007m1 2009m1period
_b[hml] lower/upper
Smallest firms, lowest value
Growth Portfolio
-2-1
01
S5
L1
2001m1 2003m1 2005m1 2007m1 2009m1period
_b[hml] lower/upper
Largest firms, lowest value
Growth Portfolio
-2-1
01
23
S1
L5
2001m1 2003m1 2005m1 2007m1 2009m1period
_b[hml] lower/upper
Smallest firms, highest value
Value Portfolio
-10
-50
5
S5
L5
2001m1 2003m1 2005m1 2007m1 2009m1period
_b[hml] lower/upper
Largest firms, highest value
Value Portfolio
-4-2
02
4
S1
L1
2001m1 2003m1 2005m1 2007m1 2009m1period
_b[hml] lower/upper
Smallest firms, lowest value
Growth Portfolio
-.5
0.5
1
S5
L1
2001m1 2003m1 2005m1 2007m1 2009m1period
_b[hml] lower/upper
Largest firms, lowest value
Growth Portfolio
-.5
0.5
11.5
S1
L5
2001m1 2003m1 2005m1 2007m1 2009m1period
_b[hml] lower/upper
Smallest firms, highest value
Value Portfolio
-2-1
01
23
S5
L5
2001m1 2003m1 2005m1 2007m1 2009m1period
_b[hml] lower/upper
Largest firms, highest value
Value Portfolio
34
1c. Patterns of HML for Portfolios sorted on Size and BE/ME: India
1d. Patterns of HML for Portfolios sorted on Size and BE/ME: China
-20
24
68
S1
L1
2001m1 2003m1 2005m1 2007m1 2009m1period
_b[hml] lower/upper
Smallest firms, lowest value
Growth Portfolio
-10
12
34
S5
L1
2001m1 2003m1 2005m1 2007m1 2009m1period
_b[hml] lower/upper
Largest firms, lowest value
Growth Portfolio-1
01
23
S1
L5
2001m1 2003m1 2005m1 2007m1 2009m1period
_b[hml] lower/upper
Smallest firms, highest value
Value Portfolio
-4-2
02
4
S5
L5
2001m1 2003m1 2005m1 2007m1 2009m1period
_b[hml] lower/upper
Largest firms, highest value
Value Portfolio
-1.5
-1-.
50
.5
S1
L1
2001m1 2003m1 2005m1 2007m1 2009m1period
_b[hml] lower/upper
Smallest firms, lowest value
Growth Portfolio
-2-1
01
2
S5
L1
2001m1 2003m1 2005m1 2007m1 2009m1period
_b[hml] lower/upper
Largest firms, lowest value
Growth Portfolio
-20
24
S1
L5
2001m1 2003m1 2005m1 2007m1 2009m1period
_b[hml] lower/upper
Smallest firms, highest value
Value Portfolio
-.5
0.5
1
S5
L5
2001m1 2003m1 2005m1 2007m1 2009m1period
_b[hml] lower/upper
Largest firms, highest value
Value Portfolio
35
Appendix
Table A1: Portfolio composition - numbers of firms: by country and year
Portfolio
BRAZIL
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Total
BG 23 30 30 30 31 32 29 36 35 44 320
BM 21 30 31 31 32 32 38 34 36 40 325
BV 23 8 4 10 8 7 4 8 14 10 96
SG 8 8 7 8 7 7 11 7 12 7 82
SM 20 20 16 15 16 18 18 24 25 31 203
SV 29 37 42 41 46 46 45 46 53 56 441
Total 124 133 130 135 140 142 145 155 175 188 1467
Portfolio
TURKEY
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Total
BG 18 21 22 25 23 26 28 28 32 32 255
BM 13 16 21 22 23 27 22 25 32 35 236
BV 7 6 6 10 13 9 13 15 14 12 105
SG 5 6 8 10 13 12 11 13 16 12 105
SM 17 17 17 22 24 22 27 29 29 27 231
SV 16 21 24 25 23 29 26 26 34 36 260
Total 76 87 98 114 119 125 127 136 157 158 1197
Portfolio
INDIA
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Total
BG 36 41 40 43 44 47 57 58 63 67 496
BM 27 26 35 39 48 44 54 55 56 63 447
BV 3 7 5 6 6 9 14 18 17 13 98
SG 4 4 9 11 15 14 18 21 19 19 134
SM 26 32 28 30 30 35 46 50 53 51 381
SV 37 38 44 48 53 52 61 61 65 73 532
Total 133 148 161 177 196 201 250 263 273 286 2088
Portfolio
CHINA
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Total
BG 81 94 106 111 119 136 158 170 205 208 1388
BM 131 149 160 174 179 213 219 215 213 270 1923
BV 103 115 157 171 190 165 177 167 165 154 1564
SG 108 121 149 163 175 173 175 162 146 172 1544
SM 121 137 177 190 210 198 224 226 252 235 1970
SV 86 100 98 103 104 144 156 165 186 226 1368
Total 630 716 847 912 977 1029 1109 1105 1167 1265 9757
36
Table A2: Simple Monthly Excess Returns for Portfolios Sorted on Size and Value
- 7/1999-6/2009, 120 Months (%)
We construct 25 portfolios by intersecting size (BE) and value (BE/ME) using quintile breakpoints. For instance
S1L1 refers to the lowest quintiles (bottom 20%) in size and value. We generate returns using a three factor
model: Rpt - Rft = α + b (Rmt-Rft) + s SMB + h HML + ε (Fama and French, 1993).
*,**,*** indicate statistical significance at the 10, 5, and 1 percentage levels.
Brazil
Value Low 2 3 4 High
Size α
Small 3.18* 0.26 7.54* 3.03** 4.45*** 2 0.31 2.75** 1.95** 3.11*** 2.86*** 3 0.86 2.54** 2.92** 1.65*** 3.14* 4 2.57*** 1.74* 2.10** 2.60*** 4.56*** Big 1.56** 1.91** 3.16*** 2.66*** 2.54
b
Small 0.05 0.06 0.49 0.15 0.12 2 0.17 0.17 0.32 0.18 0.21
3 0.29*** 0.1 0.17 0.20* 0.05
4 0.15 0.35*** 0.24 0.21 -0.09
Big 0.19 0.13 0.18 0.19 1.49
s
Small 1.01* 0.29 -0.56 0.14 0.52** 2 0.43 0.70 0.09 0.42 -0.39*
3 0.12 -0.02 0.48 0.21 -0.19
4 0.03 0.26** -0.23 -0.34* -0.77***
Big -0.12 -0.53** -0.73*** -0.75*** 2.64**
h
Small
-0.49 -0.30 0.29 -0.04 0.13 2 -0.46 0.24 0.15 0.01 0.71***
3 0.08 0.01 -0.21 0.01 0.97***
4 -0.15 -0.14 0.18 0.26* 0.82***
Big -0.05 0.23 0.34* 0.43*** -1.85
Turkey
Size α
Small -1.68 0.65 -0.75 -0.11 0.04 2 0.23 -0.91 -0.65 -0.50 -0.35
3 1.40 -0.64 -1.12*** -0.94 0.37
4 -1.61** -1.24*** -1.84*** 0.26 -0.19
Big -0.72 -1.54** -2.08*** -1.26 1.38
b
Small 0.46** 0.74*** 0.34*** 0.54*** 0.55*** 2 1.49* 0.61*** 0.58*** 0.65*** 0.36***
3 0.94*** 0.57** 0.45*** 0.47*** 0.75***
4 0.46*** 0.34*** 0.23* 0.34*** 0.45***
Big 0.61*** 0.36** 0.08 0.41*** 0.44**
s
Small 1.45*** 0.89*** 1.04*** 1.16*** 1.13*** 2 2.54** 1.17*** 0.93*** 1.29*** 1.34***
3 1.45*** 0.78** 1.14*** 1.15*** 1.37***
4 0.99*** 0.94*** 1.12*** 0.97*** 0.92***
Big 0.37** 0.94*** 0.89*** 0.92*** 0.29
h
Small
0.51*** 0.04 0.54*** 0.37*** 0.44*** 2 0.04 0.42*** 0.41*** 0.36*** 0.66***
3 -0.11 0.30 0.55*** 0.41*** 0.51***
4 0.53*** 0.63*** 0.77*** 0.64*** 0.55***
Big 0.25 0.72*** 0.98*** 0.70*** 0.57***
37
India
Value Low 2 3 4 High
Size α
Small 2.14 2.65* 2.33* 3.06** 3.28*** 2 2.46 2.29* 2.46 2.66* 3.78***
3 3.84 1.62* 2.18 2.25** 3.55**
4 1.78 2.13 1.95* 3.53** 1.37
Big 0.98 1.85 1.76 1.95 4.30***
b
Small 0.26 0.31** 0.32** 0.33** 0.33** 2 0.55*** 0.41** 0.55 0.47*** 0.44**
3 0.13 0.56*** 0.41*** 0.35*** 0.42***
4 0.28* 0.40** 0.35** 0.39** 0.33**
Big 0.46*** 0.43** 0.52*** 0.47*** 0.51***
s
Small -0.01 0.06 0.12 -0.03 -0.01 2 0.24 -0.02 0.24 -0.05 -0.08
3 0.05 0.00 0.04 0.01 0.06
4 -0.02 0.04 0.04 0.06 0.08
Big 0.08 0.04 0.05 0.01 0.06
h
Small
0.01 0.04 0.09 0.07* 0.06 2 0.22 0.01 0.22 0.03 -0.04
3 0.04 0.79 0.05 0.05 0.11**
4 0.08 0.03 0.12*** 0.11* 0.08
Big 0.06 0.10* 0.08 0.08 0.02
China
Size α
Small 0.56*** 0.91*** 1.55** 1.95*** 1.06** 2 -0.02 1.18** 1.34** 1.70*** 1.38**
3 0.17 0.14 0.31 0.96* 0.30
4 -0.22 0.35 0.23 0.42** 0.20
Big -0.10 0.22 0.15 0.24 0.24*
b
Small 0.85*** 0.90*** 0.67*** 0.62*** 1.14*** 2 1.02*** 0.89*** 0.73*** 0.58*** 0.75***
3 0.74*** 0.81*** 0.97*** 0.80*** 0.96***
4 0.99*** 0.97*** 0.79*** 1.04*** 1.02***
Big 0.72*** 0.93*** 0.92*** 0.96*** 1.04***
s
Small 0.93*** 0.88*** 0.32 -0.17 0.99*** 2 0.86*** 0.47** 0.21 0.16 0.21
3 0.79*** 0.84*** 0.71*** 0.28 0.78***
4 0.66*** 0.26*** 0.65*** 0.68*** 0.64***
Big 0.20* 0.17** 0.11** 0.12** 0.06
h
Small
-0.32*** -0.30** -0.25 -1.15*** 0.28 2 -0.29*** -0.70*** -0.55** -0.79* -0.42
3 -0.11 0.21 0.00 -0.46* 0.32**
4 -0.34*** -0.30** 0.14 0.07 0.41***
Big -0.26 -0.36*** -0.03 0.20*** 0.48***
38
Table A3: Simple Monthly Excess Returns on Portfolios Sorted on Size and Value
– July 1999 to June 2009, 120 Months: Summary Statistics (%)
We construct 25 portfolios by intersecting size (BE) and value (BE/ME) using quintile breakpoints.
For instance S1L1 refers to the lowest quintiles (bottom 20%) in size and value. PT = portfolio and
RPTRFT = return on portfolio minus the risk-free rate. We generate returns using a three factor
model: Rpt - Rft = α + b (Rmt-Rft) + s SMB + h HML + ε (Fama and French, 1993).
We test Ho = 0; H1 > 0. *,**,*** indicate significance at the 10, 5, and 1 percentage levels.
PT RPTRFT PT RPTRFT PT RPTRFT PT RPTRFT PT RPTRFT
Brazil
S1L1 3.59** S2L1 0.62 S3L1 1.69** S4L1 2.81*** S5L1 1.90***
S1L2 0.26 S2L2 4.02*** S3L2 2.78*** S4L2 2.69*** S5L2 2.07***
S1L3 8.97*** S2L3 3.02*** S3L3 3.52*** S4L3 2.72*** S5L3 3.39***
S1L4 3.48*** S2L4 3.93*** S3L4 2.34*** S4L4 3.15*** S5L4 3.00***
S1L5 5.30*** S2L5 3.85*** S3L5 4.16*** S4L5 4.35*** S5L5 0.46
Turkey
S1L1 0.58 S2L1 2.18 S3L1 2.21 S4L1 0.26 S5L1 -0.60
S1L2 1.31 S2L2 0.84 S3L2 0.53 S4L2 0.84 S5L2 0.71
S1L3 1.22 S2L3 0.74 S3L3 0.94 S4L3 0.73 S5L3 0.75
S1L4 1.56 S2L4 1.24 S3L4 0.84 S4L4 2.39** S5L4 0.89
S1L5 1.82* S2L5 2.15* S3L5 2.47* S4L5 0.76 S5L5 3.54***
India
INDIA S1L1 2.51* S2L1 4.18 S3L1 4.21** S4L1 2.26** S5L1 1.94**
S1L2 3.31** S2L2 2.84** S3L2 2.53** S4L2 2.84*** S5L2 2.71**
S1L3 3.22*** S2L3 2.82** S3L3 2.94** S4L3 2.73** S5L3 2.75**
S1L4 3.56*** S2L4 3.24*** S3L4 2.84** S4L4 4.39*** S5L4 2.83**
S1L5 3.82*** S2L5 4.15*** S3L5 4.47*** S4L5 2.16** S5L5 5.38***
China
CHINA S1L1 1.31** S2L1 0.83 S3L1 0.93* S4L1 0.49 S5L1 0.27
S1L2 1.69*** S2L2 1.51** S3L2 1.17** S4L2 0.87* S5L2 0.65
S1L3 2.00*** S2L3 1.54** S3L3 1.23** S4L3 1.11** S5L3 0.74
S1L4 1.53** S2L4 1.63** S3L4 1.29** S4L4 1.43** S5L4 1.01*
S1L5 2.40*** S2L5 1.67** S3L5 1.46** S4L5 1.39** S5L5 1.20**
39
Table A4: Mean Coefficients of Rolling Regressions: Portfolios sorted on Size and Value
See Table A2. We estimate coefficients on 36 month rolling regressions of a three factor model: Rpt - Rft = α + b
(Rmt-Rft) + s SMB + h HML + ε. Rm - Rf is the return on the market portfolio less the risk-free rate. SMB is the
difference between the simple average of returns on small stock (S/L, S/M, S/H) and big stock portfolios (B/L,
B/M, B/H). HML is the difference between the simple average of returns on high value (S/H, B/H) and low
value portfolios (S/L, B/L).
L1 L2 L3 L4 L5 L1 L2 L3 L4 L5
Brazil
Constant Market
S1 4.810 -0.544 4.752 3.298 3.740 0.097 0.194 0.296 0.212 0.312
S2 -0.179 3.387 1.571 3.135 2.802 0.211 0.321 0.361 0.124 0.570
S3 0.249 2.432 1.921 1.056 3.471 0.386 0.144 0.278 0.325 0.155
S4 2.220 1.129 1.890 2.007 2.078 0.001 0.256 0.225 0.313 -0.620
S5 1.257 1.066 2.194 1.715 4.023 0.372 0.317 0.526 0.503 1.534
SMB HML
S1 1.742 0.309 -2.341 0.382 0.841 -0.996 -0.297 1.157 -0.228 -0.046
S2 0.760 0.951 0.087 0.464 -0.257 -0.769 -0.080 0.087 -0.081 0.777
S3 -0.030 0.165 0.024 -0.063 -0.106 0.275 -0.145 0.207 0.288 0.809
S4 -0.242 0.219 -0.304 -0.320 -0.513 0.100 -0.121 0.166 0.284 0.462
S5 -0.158 -0.457 -0.700 -0.474 3.483 0.002 0.259 0.369 0.363 -2.465
Turkey
Constant Market
S1 -1.212 1.174 -0.760 -0.389 -0.059 0.025 0.733 0.390 0.453 0.352
S2 -2.760 -0.676 -0.783 -1.238 -0.209 0.494 0.644 0.226 0.471 0.262
S3 3.367 -1.641 -1.012 -1.506 0.126 1.320 0.239 0.460 0.576 0.671
S4 -2.311 -1.407 -2.324 -0.132 -0.533 0.174 0.320 0.144 0.390 0.631
S5 -0.822 -1.396 -2.300 -0.862 1.062 0.727 0.311 0.295 0.552 0.583
SMB HML
S1 1.347 0.494 1.108 1.096 1.110 0.771 -0.085 0.552 0.459 0.541
S2 1.935 0.920 0.959 1.107 1.321 0.655 0.397 0.696 0.514 0.692
S3 1.995 1.194 1.031 1.020 1.255 -0.326 0.774 0.518 0.392 0.547
S4 0.917 0.957 1.142 1.008 0.919 0.637 0.654 0.894 0.670 0.499
S5 0.311 0.708 0.803 0.396 -0.062 0.186 0.662 0.838 0.418 0.336
India
Constant Market
S1 1.407 1.820 1.594 1.590 2.194 -0.005 0.407 0.182 0.206 0.179
S2 -0.411 1.258 1.038 0.988 2.835 0.263 0.282 0.332 0.336 0.220
S3 5.427 0.507 1.177 1.010 2.425 -0.322 0.362 0.327 0.267 0.238
S4 0.115 0.896 1.003 2.400 0.138 0.147 0.225 0.157 0.248 -0.050
S5 0.189 0.748 -0.023 0.681 3.002 0.291 0.275 0.476 0.424 0.578
SMB HML
S1 -0.690 -0.075 -0.745 -0.601 -0.519 0.573 0.171 0.695 0.505 0.416
S2 -0.545 -0.669 -0.571 -0.644 -0.705 0.661 0.414 0.497 0.524 0.426
S3 -0.455 -0.689 -0.670 -0.697 -0.773 0.329 0.645 0.438 0.507 0.731
S4 -0.643 -0.600 -0.654 -0.711 -1.576 0.616 0.489 0.623 0.615 0.672
S5 -0.448 -0.588 -0.632 -0.525 -0.287 0.438 0.519 0.560 0.406 0.079
China
Constant Market
S1 0.608 0.706 0.505 0.413 0.461 0.852 0.847 0.595 0.468 1.119
S2 -0.191 0.619 0.686 0.325 1.808 1.021 0.751 0.398 0.418 0.844
S3 0.252 0.467 0.053 -0.127 0.300 0.882 0.873 0.843 0.674 0.874
S4 0.158 0.486 0.671 0.456 0.353 1.063 0.997 0.865 1.021 1.032
S5 0.347 0.489 0.467 0.379 0.425 0.908 0.949 0.967 0.972 1.030
SMB HML
S1 0.842 0.724 -0.023 0.019 1.199 -0.409 -0.406 -0.177 -0.440 0.317
S2 0.791 0.623 0.314 0.473 0.391 -0.312 -0.352 -0.355 0.021 -0.232
S3 0.633 0.781 0.548 0.523 0.654 -0.131 0.108 0.002 0.037 0.328
S4 0.498 0.243 0.283 0.512 0.493 -0.662 -0.385 -0.057 0.133 0.280
S5 -0.230 -0.122 -0.064 0.002 -0.060 -0.562 -0.543 -0.101 0.096 0.364
40